Hindawi Publishing CorporationJournal of OphthalmologyVolume 2010, Article ID 746978, 10 pagesdoi:10.1155/2010/746978

Research Article

NADPH Oxidase versus Mitochondria-DerivedROS in Glucose-Induced Apoptosis of Pericytes inEarly Diabetic Retinopathy

Nik M. Mustapha,1, 2 Joanna M. Tarr,3 Eva M. Kohner,3 and Rakesh Chibber2, 3

1 Forest Research Institute Malaysia (FRIM), 52109 Kepong, Selangor Darul Ehsan, Malaysia2 Cardiovascular Division, GKT School of Biomedical & Health Sciences, King’s College London, Guy’s Campus,London SE1 1UL, UK

3 Institute of Biomedical and Clinical Science, Peninsula College of Medicine and Dentistry, Peninsula Medical School,St Luke’s Campus, Exeter EX1 2LU, UK

Correspondence should be addressed to Rakesh Chibber, rakesh.chibber@pms.ac.uk

Received 1 January 2010; Revised 29 March 2010; Accepted 23 April 2010

Academic Editor: Renu A. Kowluru

Copyright © 2010 Nik M. Mustapha et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Objectives. Using apocynin (inhibitor of NADPH oxidase), and Mitoquinol 10 nitrate (MitoQ; mitochondrial-targeted antiox-idant), we addressed the importance of mitochondria versus NADPH oxidase-derived ROS in glucose-induced apoptosis ofpericytes. Methods. NADPH oxidase was localised using Western blot analysis and cytochrome C reduction assay. Apoptosiswas detected by measuring caspase-3 activity. Intracellular glucose concentration, ROS formation and Nε-(carboxymethyl) lysine(CML) content were measured using Amplex Red assay kit, dihydroethidium (DHE), and competitive immunoabsorbant enzyme-linked assay (ELISA), respectively. Results. NADPH oxidase was localised in the cytoplasm of pericytes suggesting ROS productionwithin intracellular compartments. High glucose (25 mM) significantly increased apoptosis, intracellular glucose concentration,and CML content. Apoptosis was associated with increased gp91phox expression, activity of NADPH oxidase, and intracellularROS production. Apocynin and not MitoQ significantly blunted the generation of ROS, formation of intracellular CML andapoptosis. Conclusions. NADPH oxidase and not mitochondria-derived ROS is responsible for the accelerated apoptosis ofpericytes in diabetic retinopathy.

1. Introduction

Diabetic retinopathy is a leading cause of blindness that ischaracterized by vascular changes of the retinal capillary bed[1]. One of earliest changes is the accelerated apoptosis ofretinal microvascular cells and the formation of acellularcapillaries [2]. Although, the frequency of pericytes andendothelial cell apoptosis is thought to predict the devel-opment of the histological lesions in retinopathy [3], theunderlying cause is not fully understood. Diabetes Controland Complications Trial (DCCT) and the United KingdomProspective Diabetes Study (UKPDS) have confirmed thathyperglycaemia is the major factor in the development ofdiabetic retinopathy [4, 5]. In addition to the prevailing

biochemical mechanisms [1] on how glucose leads topathological changes in retinopathy, recent evidence suggestsa key role for oxidative stress, a state in which excess reactiveoxygen species (ROS) overwhelm endogenous antioxidantsystems [6]. ROS can be produced from the mitochondrialtransport chain and a number of enzymes that are localisedin the plasma membrane, and the cytoplasm of cells [7].

The elegant study carried out by Brownlee and colleagues[8] suggests that glucose-induced mitochondria productionof ROS stimulates several of the biochemical mechanismthought to be involved in hyperglycaemia-mediated com-plications of diabetes, including retinopathy. The authorsproposed that the causal link between glucose and vasculardamage in diabetes is the increased production of superoxide

brought to you by COREView metadata, citation and similar papers at core.ac.uk

provided by PubMed Central

2 Journal of Ophthalmology

by the mitochondrial electron transport chain [8]. However,increasing evidence suggests that NADPH oxidase is themost important source of cellular ROS in blood vessels[9]. NADPH oxidase complex in neutrophils and, mostprobably, endothelial cells and other cell types, involves fouressential subunits. The subunits gp91phox and p22phoxreside in the plasma membrane [9]. These subunits bind thecomponents of the electron transport chain heme and FAD,forming cytochrome b558. The cytosolic NADPH oxidasesubunits p47phox and p67phox are involved in the activationof the enzyme complex. Unlike the phagocytic type, theNADPH oxidases present in blood vessels are constitutivelyactive, producing relatively low levels of ROS under basalconditions, and generating higher levels of oxidants inresponse to cytokines [9]. Among the nonphagocytic cellsexamined so far, endothelial NADPH oxidase has beeninvestigated more extensively [9].

By using apocynin and Mitoquinol 10 nitrate (MitoQ) weaddressed the importance of mitochondria versus NADPHoxidase-derived ROS in glucose-induced apoptosis of cul-tured retinal capillary pericytes. Apocynin is a methoxy-substituted catechol that does not act as a ROS scavenger,but inhibits NADPH oxidase by impeding the assemblyof p47phox and p67phox subunits within the membraneNADPH oxidase complex [10]. MitoQ which has an antioxi-dant ubiquinol moiety attached to a triphenylphosphoniumcation by an aliphatic carbon chain is recently developedmitochondria-targeted antioxidants that selectively blockmitochondrial oxidative damage [11]. The selective accumu-lation of MitoQ prevents mitochondrial oxidative damagefar more effectively than untargeted antioxidants. It is notonly accumulated by the mitochondria but also can beregenerated in its reduced form by mitochondrial respiratorychain [11]. Here we show for the first time, that it is ROSderived from NADPH oxidase and not the mitochondria,which is involved in apoptosis of pericytes induced bychronic exposure to high glucose.

2. Materials and Methods

2.1. Culture of Bovine Retinal Capillary Pericytes. Bovine reti-nal capillary pericytes (BRPs) were established from bovineretinas dissected from eyes of freshly slaughtered cattle asdescribed previously in [12]. Briefly, the isolated retinaswere homogenised in serum-free minimal essential medium(MEM; Sigma, UK), and filtered through 80 μm nylon mesh.The trapped micro vessels were digested with collagenase-dispase (1 mg/ml) for 30 min at 37◦C, filtered through a45 μm nylon mesh and then plated in tissue culture flasksand maintained in MEM supplemented with 10% foetalcalf serum (FCS), 2 mM glutamine, 100 IU/ml penicillin,and 100 μg/ml streptomycin. The cells were characterized asdescribed previously [12] and used at passage 2-3.

For experiments, confluent cultures of BRP were exposedto continuous normal (5.6 mM) glucose and high (25 mM)glucose for 4 days. In some experiments, 500 μM apocyninand 1 μM of the mitochondrial targeted antioxidant [11],MitoQ (quinol attached to triphenylphosphonium; kindly

provided by Dr Murphy, MRC-Dunn Human NutritionUnit, Cambridge, UK) was added to normal and high-glucose medium.

2.2. Caspase-3-Like Activity. Cellular caspase-3 activity wasdetermined using the colorimetric protease assay. This assaydetects p-nitroanilide (p-NA) photometrically at 405 nmafter its cleavage from the colorimetric substrate, N-acetyl-Asp-Glu-Val p-nitroanilide (Ac-DEVD-pNA) by the caspase-3 [13].

2.3. DNA Fragmentation. DNA fragmentation was quanti-fied using the Cell Death Detection ELISA plus (1774425;Roche, Hertfordshire, UK). The assay was carried outaccording to manufacturer’s instructions, which measuresmono-, and oligonucleosomes in the cytoplasmic fractionof cell lysate. The assay was based on a quantitive sandwichenzyme-immunoassay directed against cytoplasmic histone-associated DNA fragments.

2.4. Western Blot Analysis. Antibodies directed againstNADPH oxidase subunits, Gp91phox, and p47phox werefrom Upstate (UK). After treatment, cells were lysed onice in the following lysis buffer (20 mM Tris-HCL, pH7.4, 1% Triton X-100, 150 mM NaCl, 1 mM EDTA, 1 mMEGTA, 0.2 mM sodium vanadate, 1 mM PMSF, 1 μg/mlaprotinin, 10 μg/ml leupeptin). Equal amounts of proteinwere separated on 10% polyacrylamide gels and transferredto Hybond-ECL membranes (Amersham, UK). Membraneswere blocked with 3% bovine serum albumin (BSA) inphosphate buffered saline (PBS) containing 0.05% Tween 20(PBST). After immunodetection of p47phox, and gp91phox,the membranes were stripped in a buffer containing50 mM Tris, 2% SDS, and 100 mM mercaptoethanol at55◦C for 30 min, washed and immunoblotted with tubulinantibody (Chemicon, Hampshire, UK). Immunoreactivebands were quantified by scanning densitometrically andcalculating the density of individual bands using Image Jsoftware (National Institute of Health, Bethesda, Maryland,http://rsb.info.nih.gov/ij/). The levels of NADPH oxidasesubunits were expressed as a ratio of intensity of p47phox,and gp91phox immunoreactive bands/intensity of tubulinimmunoreactive bands.

2.5. Measurement of Intracellular ROS Production. The cellpermeant dihydroethidium (DHE; Molecular Probes) wasused to assess real-time formation of superoxide (O2

−)in BRP exposed to normal (5.8 mM) and high (25 mM)glucose. DHE enters the cells and is oxidized by superoxide toform ethidium (ETH), which binds to DNA to produce thefluorescent ETH-DNA that displays red fluorescence [14].DHE was prepared as 2 mM stock solution in DMSO andstored at −20◦C. The cells were loaded with 10 μM for thefinal 2 h of incubation. The cells were washed with PBS,lysed with 50 μl lysis buffer (20 mM Tris-HCl, pH 7.4, 1%Triton X-100, 150 mM NaCl, 1 mM EDTA, 50 mM NaF and0.2 mM sodium orthovanadate) on ice. The lysates were thentransferred into black 96-well plates (Fisher, Loughborough,

Journal of Ophthalmology 3

UK) and the fluorescence was measured using spectrofluo-rometer (Plate Chameleon; Hidex, Baringstoke, UK). ETH-DNA red fluorescence was measured with excitation at530 nm and emission at 616 nm. Background fluorescenceintensity was subtracted from the results and the level of ROSwas expressed as fluorescence intensity/μg protein.

2.6. Measurement of NADPH Oxidase Activity. Measurementof O2

− was based on the capacity to reduce ferricytochromec in ferrocytochrome at pH 7.8 [15]. Total cell lysate(50 μg protein/experiment), cytochrome c (250 μg/l finalconcentration), and NADPH (100 μM) were incubated at37◦C for 120 min, either in the presence or absence ofdiphenyleneiodonium (DPI, 100 μM). The reduction ofcytochrome c was measured by reading absorbance at550 nm. O2

− production in nmol/mg protein was calculatedfrom the difference between absorbance of samples at 0 and120 min and the extinction coefficient 21 mmol/l/cm.

2.7. Measurement of Intracellular CML. The intracellularformation of Nε-(carboxymethyl) lysine (CML) in pericyteswas quantified using competitive imuunoabsorbant enzyme-linked assay (ELISA) as previously described in [16]. CML-BSA, used as a standard, was dissolved in 0.05 M carbonatebuffer, pH 9.6, to a concentration of 0.5 μg/ml. A 50 μlaliquot was added to each well of a 96-well microtitreELISA plate (Nunc, Loughborough, UK). After incubationat 4◦C overnight, the coating solution was discarded andthe wells washed three times with 400 μl of PBS containing0.05% Tween-20 (PBST). The wells were then filled with100 μl of 2% normal goat serum (NGS) containing 0.1%BSA for blocking and left for 1 h at room temperature.After washing with three times with 400 μl PBST, 50 μl ofthe diluted standards and samples were added in triplicatewells followed by 50 μl of the 1 : 1000 diluted CML-antibody(Kindly provided by Dr D Ruggiero, INSA, Lyon, France).After incubation for 2 h at room temperature, the wells werewashed three times with 400 μl PBST, and developed withgoat antirabbit peroxidase-conjugated IgG (diluted 1 : 10,000in PBST) and o-phenylenediamine (Sigma, Dorset, UK). Theabsorbance of the samples was read at 490 nm and the levelsof CML determined using the standard curve prepared usingvarious concentrations of CML-BSA.

2.8. Measurement of Intracellular Glucose Concentration.After treatment, cells were washed and medium was removedand cells lysed. Intracellular glucose concentration was thendetermined using Amplex Red Glucose Assay Kit (A 22189;Molecular Probes, Paisley, UK) according to the manufac-turer’s instructions. Briefly, 50 μl of the reaction solution(10 mM Amplex Red, 10 U/ml HRP, 100 U/ml glucose oxi-dase, 50 mM sodium phosphate buffer, pH 7.4) was added to50 μl of cell lysate in 96-well microtitre plate and incubatedin the dark for 45 min at room temperature. The absorbancewas then measured at 560 nm using a SpectraMax 190microplate reader. Intracellular glucose concentration wasdetermined from a standard curve generated using variousconcentrations of glucose.

2.9. Measurement of Protein Kinase Cβ1/2 Activity. Cellu-lar PKCβ1/2 activity was determined using the TruLightassay kit (Calbiochem, Nottingham, UK) according to themanufacturer’s instructions. The activity was based onfluorescence superquenching and the assay was carried outin a white 96-well microtitre plates (Nunc, Loughborough,UK). Fluorescence was measured using microplate reader(Plate Chameleon, Hidex, Basingstoke, UK) with excitationat 450 nm and emission at 535 nm. PKCβ1/2 activity wascalculated from the generated phosphopeptide calibratorcurve and the results expressed PKCβ1/2-dependent phos-phorylation/mg protein.

2.10. Protein Measurement. Total protein was measuredusing the BCA protein assay kit (Pierce, UK).

2.11. Statistical Analysis. The statistical software Graph PadPrism version 3.0 was used. A two-tailed Student’s t-testwas used to test the significance of paired data. Data areexpressed as means± S.E.M of measurements in the differentexperiments. Differences between groups were consideredstatistically significant at P < .05.

3. Results

3.1. Retinal Capillary Pericytes Express NADPH Oxidase.NADPH oxidase is a membrane-bound enzyme complexthat is a source of cytosolic O2

− and is composed of atleast four subunits, including two important membrane-bound subunits p22phox and gp91phox and two cytosolicsubunits p47phox and p67phox. On a Western blot, theantibody directed against the N-terminal peptide labelledspecifically only one broad band at approximately 80 kDain lysates of BRP (Figure 1(a)). This immunoreactive bandwas detected using the antibody directed against the peptidecorresponding to amino acids 548–560 of human gp91phox.The antibody also detected similar molecular weight bandin cultured bovine retinal capillary endothelial cells (BREC),human umbilical endothelial cells (HUVEC), and humanleukocytes (U937 cells) (Figure 1(a)). The anti-gp91phoxantibody used in our study is specific to human gp91phox(Nox2) and corresponds to amino acids 548–560 of humanNox2.

Specific anti-p47phox also confirmed the presence ofp47phox NADPH oxidase subunit with approximate molec-ular weight of 55 kDa in BRP. An immunoreactive band ofsimilar weight was also localized in BREC, and HUVEC (Fig-ure 1(a)). However, in U937 cells the p47phox immunore-active band had a lower molecular weight of approximately45 kDa. Thus, two major components of a functionalNADPH oxidase are present in BRP.

Immunoblot analysis showed that 48 h exposure to highglucose significantly increased gp91phox protein expression(Figures 1(b) and 1(c)) to 249 ± 38% of normal glucose(n = 5, P < .05). In contrast, high glucose failed to causea significant change in the expression of p47phox (Figures1(b) and 1(d)).

4 Journal of Ophthalmology

gp91phox

p47phox

1 2 3 4

80 kDa

55 kDa

∼

∼

(a)

NG HG

gp91phox

p47phox

Tubulin

(b)

0

100

200

300

400

∗

NG HG

Rel

ativ

ein

ten

sity

(%of

con

trol

)

(c)

0

1 0 0

2 0 0

3 0 0

4 0 0

N G H G

R e l

a t i v

e i n

t e n

s i t y

( %

o f c

o n t r

o l )

(d)

Figure 1: Retinal capillary pericytes express NADPH oxidase. Aliquots of total cellular protein were separated by SDS-PAGE, electrophoret-ically transferred onto Hybond-ECL membranes. Blots were probed with (a) anti-gp91phox antibody and anti-p47phox antibody. Proteinantibody complexes were detected using secondary antibody conjugated with horseradish peroxidase and the ECL detection system. Lane1: BRP, lane 2: BREC (bovine retinal capillary endothelial cells), Lane 3: HUVEC (human umbilical vein endothelial cells), and lane 4:U937 cells (human leukocytes). (b) Representative Western blots of gp91phox, p47phox and tubulin. (c) Expression of gp91phox proteinexpression in BRP exposed to high glucose (HG, 25 mM) for 48 h. Densitometric ratio (intensity gp91phox immunoreactive band/intensityof the tubulin immunoreactive band) is expressed as % of normal glucose (NG, 5.6 mM). (d) Expression of p47phox protein expressionin BRP exposed to high glucose (HG, 25 mM) for 48 h. Densitometric ratio (intensity of p47phox immunoreactive band/intensity of thetubulin immunoreactive band) is expressed as % of normal glucose (NG, 5.6 mM). Data are represented as Means ± SEM of 3-4 separateexperiments. ∗P < .05 versus control.

3.2. Glucose Increases NADPH Oxidase Activity. To providefurther and more direct evidence for the effect of highglucose on NADPH oxidase, cellular activity was measuredusing the cytochrome c reduction assay. BRP expressesactive NADPH oxidase (Figure 2(a)). High glucose increasedthe NADPH-induced O2

− generation by almost 1.65-foldcompared to normal glucose (Figure 2(b)). The addition ofapocynin (500 μM) significantly inhibited glucose-inducedNADPH oxidase activity (115 ± 22% of normal glucose,n = 4, P = .013).

3.3. Glucose Stimulates ROS in Pericytes. Exposure to highglucose significantly increased the intracellular O2

− pro-duction as detected by ETH-DNA fluorescence (Figure 3).Oxidant generation increased 126 ± 9% of normal glucose(n = 6, P < .05) and was significantly reversed by treatmentwith 500 μM apocynin (101 ± 7% of normal glucose versus126 ± 9% of normal glucose, n = 6, P < .05). MitoQ

slightly reversed glucose-induced ROS production (122 ±4% of normal glucose, n = 6, P = .05).

3.4. Apocynin and Not MitoQ Reverse Glucose-InducedApoptosis. Figure 4 compares the effect of high glucoseon apoptosis determined by measuring cellular caspase-3activity (A) and the level of DNA fragmentation (B) using theCell Death Detection ELISA Plus (Roche, Hertfordshire, UK)which measures mono- and oligonucleosomes in the celllysate. Both assay confirmed high glucose-induced apoptosisof BRP after 4 days.

As shown in Figure 5, incubation for 4 days in continuoushigh glucose caused a significant increase in apoptosis ofpericytes compared to normal glucose (127 ± 9% of normalglucose, n = 7, P < .001). This glucose-induced caspase-3 activity was significantly reversed (Figure 5) by 500 μMapocynin (91 ± 7% of normal glucose versus 127 ± 9% ofnormal glucose, n = 6−13, P < .05). In contrast, the addition

Journal of Ophthalmology 5

Lysate +NADPH0

0.2

0.4

0.6

0.8

∗

∗∗

+NADPH+DPI

O2−

prod

uct

ion

(nm

ol/m

gpr

otei

n)

(a)

0

50

100

150

200

250

∗

∗∗

NA

DP

Hox

idas

eac

tivi

ty(%

ofco

ntr

ol)

NG HG HGA

(b)

Figure 2: Apocynin reverses glucose-induced NADPH oxidase activity. (a) Cultured bovine retinal capillary pericytes express active NADPHoxidase. The activity was determined by cytochrome c reduction. Lysate (25 μg protein) was added to 250 μM cytochrome c in a 96-well multiwell plate and incubated for 120 min, either in the presence or absence of NADPH (100 μM) or NADPH oxidase inhibitoror flavoprotein inhibitor, diphenyleneiodonium (DPI, 100 μM). The reduction of cytochrome c was measured by reading absorbance at550 nm (SpectraMax 190). Superoxide (O2

−) production in nmol/mg protein was calculated from absorbance of samples and the extinctioncoefficient for change of ferricytochrome c to ferrocytochrome c of 21 mmol/l/cm. Data are presented as mean ± SEM of 4–6 separateexperiments. ∗P < .001 versus lysate alone, ∗∗P < .01 versus lysate with NADPH. (b) Exposure to high glucose for 4 days increases NADPHoxidase which is significantly reversed by the addition of apocynin. Confluent cultures of BRP in 30 mm2 dishes were exposed to normalglucose (NG, 5.6 mM) and high glucose (HG, 25 mM) with and without 500 μM apocynin (HGA) for 4 days. After incubation, the cells werewashed twice with ice-cold PBS; lysed, and NADPH-dependent O2

− production was measured by DPI-inhibitable cytochrome c reductionassay. Data are presented as mean ± SEM of 4 separate experiments. ∗P < .05 versus HG. ∗∗P < .013 versus HG.

NG HG HGA HGMQ0

50

100

150

FLin

ten

sity

/µg

prot

ein

(%of

con

trol

) ∗

∗∗∗∗

Figure 3: Apocynin reverses continuous high glucose-induced ROSproduction in BRP. Subconfluent cells in 24-well plate were exposedto normal glucose (NG, 5.6 mM) and continuous high glucose (HG,25 mM) in the absence and presence of 500 μM apocynin (HGA)or 1 μM MitoQ (HGMQ) for 3 days. Approximately, 2 h before theend of incubation, DHE (10 μM) was added. After the treatment,the cells were washed twice with PBS, lysed and fluorescence (FL)intensity was measured. Data are presented as mean ± SEM of 6separate experiments. ∗P < .05 versus NG, ∗∗P < .05 versus HG.

of MitoQ failed to prevent glucose-induced activation ofcaspase-3 (124 ± 17% of normal glucose, n = 4).

3.5. Intracellular CML Content after Exposure to High Glucose.Intracellular levels of CML were quantified by competitiveenzyme-linked immunoabsorbant assay (ELISA) using spe-cific CML antibody. Standard curve was established using

various concentration of CML-BSA. As shown in Figure 6,exposure to continuous high glucose for 4 days significantlyincreased the intracellular CML content by almost 2.8-foldcompared to normal glucose (16.3 ± 1.9 μg/mg proteinversus 5.9 ± 0.9 μg/mg protein in normal glucose, n =10–12, P < .05). The addition of apocynin at 500 μM,a concentration that reverses glucose-induced apoptosis,significantly prevented the accumulation of CML in pericytesexposed to high glucose (10.7 ± 1.4 μg/mg protein versus16.3± 1.9 μg/mg protein in high glucose, n = 8–12, P < .05).In contrast, MitoQ, (1 μM) had no significant effect on theCML content in pericytes exposed to high glucose (16.9 ±5.1 μg/mg protein, n = 8).

3.6. Glucose-Induced PKC-β Activity. There was no signif-icant increase in PKC-β1/β2 activity measured using fluo-rescence superquenching-based assay. After 4 days exposure,the activity expressed as PKC-β1/β2-dependent phosphory-lation/mg protein was found to be 6.9 ± 3.5 (n = 6) in HGcompared to 6.8 ± 3.5 (n = 6).

3.7. Increased Accumulation of Intracellular Glucose afterExposure to High Glucose. Continuous exposure to highglucose (25 mM) for 4 days increased the intracellularglucose concentration by almost 6-fold compared to BRPexposed to normal glucose (92.6 ± 6.0 nmol/mg proteinversus 15.2 ± 3.0 nmol/mg protein in normal glucose, n =13-14, P < .001) (Figure 7). At 500 μM, apocynin failedto prevent the accumulation of intracellular glucose (87.8± 22.3 nmol/mg protein, n = 5). Interestingly, although

6 Journal of Ophthalmology

N HG0

50

100

150

∗

Cas

pase

-3ac

tivi

ty(%

ofco

ntr

ol)

(a)

N HG0

50

100

150

200

∗

DN

Afr

agm

enta

tion

(%of

con

trol

)

(b)

Figure 4: Continuous high (25 mM) glucose induces apoptosis of BRP. Subconfluent cells in 30 mm2 dishes were exposed to normal (N,5.6 mM) and high (CG, 25 mM) glucose for 4 days. (a) High glucose activates caspase-3. After 4 days, the cells were washed twice withice-cold PBS, lysed and apoptosis was measured using colorimetric caspase activity assay. Data are represented as Means± SEM of 6 separateexperiments. ∗P = .0315 versus N. (b) High glucose induces DNA fragmentation. After 4 days, DNA fragmentation was measured usingCell Death Detection ELISA plus (Roche, Hertfordshire, UK). Data are represented as Means ± SEM of 5 separate experiments. P = .043versus N.

NG NGA NGMQ HG HGA HGMQ0

50

100

150

Cas

pase

-3ac

tivi

ty(%

ofco

ntr

ol)

∗

∗∗

Figure 5: Chronic exposure to high glucose induces apoptosis in BRP.Subconfluent cells in 3 cm dishes were exposed to normal (NG,5.6 mM) and high (HG, 25 mM) glucose in the absence (NG, HG)and presence of 500 μM apocynin (NGA, HGA) or 1 μM MitoQ(NGMQ and HGMQ). After 4 days, the cells were washed twice withice-cold PBS, lysed and apoptosis was measured using colorimetriccaspase-3 activity. Data are represented as Means ± SEM of 3–6separate experiments. ∗P < .05 versus NG, ∗∗P < .05 versus HG.

MitoQ (1 μM) failed to prevent the activation of caspase-3, it significantly lowered the level of intracellular glucoseconcentration in BRP exposed to high glucose (57.9 ±8.6 nmol/mg protein versus 92.6 ± 6.0 nmol/mg protein inhigh glucose, n = 5–13, P < .05). However, apocynin andMitoQ did not alter the intracellular glucose concentrationin BRP exposed to normal glucose.

4. Discussion

There is now growing evidence that oxidative stress playsan important role in the pathogenesis of chronic compli-cations of diabetes [6], but the exact source, and cellular

NG HG HGA HGMQ0

10

20

30

CM

L-m

odifi

edpr

otei

n(µ

g/m

gpr

otei

n)

∗

∗∗

Figure 6: Apocynin reverses glucose-induced CML formation in BRP.Subconfluent cells in 24-well plate were exposed to normal glucose(NG, 5.6 mM) and continuous high glucose (HG, 25 mM) and inthe presence of 500 μM apocynin (HGA) or 1 μM MitoQ (HGMQ)for 4 days. After incubation, the cells were washed twice with ice-cold PBS, lysed and CML were detected using competitive ELISA.Data are presented as Mean ± SEM of 8–13 samples. ∗P < .05versus NG, ∗∗P < .05 versus HG.

location of the glucose-induced ROS is still unclear. Brownleeand co-workers proposed that in cultured macrovascularbovine aortic endothelial cells (BAECs), the productionof ROS by the mitochondria via the respiratory chain isthe most important causal link between high glucose andthe main pathways responsible for hyperglycaemic damage[8]. Besides mitochondria, NADPH oxidase also generatesa significant amount of ROS and is a major source ofsuperoxide in vascular cells [9]. In the present study, we usedapocynin, an inhibitor of NADPH oxidase [10] and MitoQ,a mitochondria-targeted antioxidant [11, 12], to explorethe importance of mitochondria versus NADPH oxidase

Journal of Ophthalmology 7

NG NGA NGMQ HG HGA HGMQ0

50

100

150

Intr

acel

lula

rgl

uco

se(n

mol

/mg

prot

ein

)

∗

∗∗

Figure 7: Exposure to high glucose increases intracellular glucoseconcentration. Subconfluent cells in 24-well plate were exposed tonormal glucose (NG, 5.6 mM) and high glucose (HG, 25 mM)and in the presence of 500 μM apocynin (NGA, HGA) or 1 μMMitoQ (NGMQ, HGMQ) for 4 days. The cells were washed twicewith ice-cold PBS, lysed and intracellular glucose concentrationswere measured using the Amplex Red glucose assay kit (MolecularProbe). Data are Means ± SEM of 5–14 separate experiments. ∗P <.001 versus NG; ∗∗P < .05 versus HG.

derived ROS in glucose-induced apoptosis of cultured retinalcapillary pericytes.

Our observations of glucose-induced apoptosis of per-icytes is consistent with previous in vitro studies [17–20]and the activation of caspase-3 in the retina of diabeticanimals and humans [21, 22]. Consistent with recent reports[23, 24], retinal capillary pericytes express NADPH oxidaseas indicated by the immunoblotting of Nox2 and p47phox,major membrane and cytosolic subunits [9, 10], respectively.In contrast, Manea et al. [24] using reverse transcriptase-polymerase chain reaction (RT-PCR) detected Nox1 andNox2 in pericytes isolated from rat adipose tissue microvas-culature. Since Nox2 and Nox1 shows only 56% of homology[25], we safely assume that our antibody does not cross-react with Nox1. NADPH oxidase, as in other cell types [26,27] could be mostly present in the cytoplasm of pericytes,suggesting that ROS is produced within the intracellularcompartments. Although, glucose increased NADPH oxidasevia Nox2 expression, the exact mechanism of control is atpresent unclear. In support of our observations, increasedmRNA and/or protein levels of gp91phox have been reportedin blood vessels from animals [28, 29] and patients withdiabetes [30]. As reported in other cell types [22, 31, 32],exposure to high glucose increased NADPH oxidase activityand ROS production in pericytes. In addition to changes inthe expression of NADPH oxidase subunits [33, 34], highglucose could stimulate ROS production through proteinkinase C (PKC)-dependent phosphorylation of the p47phoxsubunit [32]. The β isoform of PKC has been implicatedin the phosphorylation of p47phox [33], but in our study,we failed to demonstrate activation of PKCβ1/2 in pericytesexposed to high glucose.

Our findings of increased oxidative stress in responseto high glucose is in line with some previous studies

[20, 23], but it appears to argue against a recent reportsuggesting that pericytes are resistant to glucose-inducedoxidative stress [35]. The reason for this inconsistency is atpresent unclear. Our results show that ROS derived fromNADPH oxidase and not the mitochondria, is involved incaspase-3-mediated apoptosis of pericytes induced by highglucose. This possibility is supported by recent studies [36]demonstrating the potential of apocynin to block glucose-induced ROS in BAEC. Other workers have demonstratedthe role of NADPH oxidase in diabetic retinopathy [36–39], and complications of diabetes [40, 41] including theloss of podocytes in diabetic nephropathy [42]. Resultswith MitoQ suggested that some ROS is produced fromthe mitochondria, but it does not play a significant role inglucose-induced apoptosis. The mitochondria is known toplay a key role in activating apoptosis through enhancedcytochrome (cyt c) release resulting in the activation ofcaspases and subsequent cell death [43]. Apoptotic celldeath is mediated by stimulated caspase-3 activity, andthe accumulation of p53, a known signalling moleculethat acts upstream of caspase-3. Although, there is nodirect evidence available to show that ROS derived fromNADPH oxidase interact with mitochondria, we suggestthat there is some interaction in pericytes exposed to highglucose.

In bovine aortic endothelial cells (BAECs) the mito-chondria-derived ROS was also involved in glucose-inducedintracellular AGE formation [6]. However, our observationthat apocynin reverses glucose-induced Nε-(carboxymethyl)lysine (CML) production, suggests a role of NADPH oxidase-derived ROS in the formation of intracellular CML-modifiedproteins in pericytes exposed to high glucose. In support ofthis, there is evidence that phagocytic NADPH oxidase playsan important role in CML formation in vivo [44]. Since CMLis a biomarker of cellular oxidative stress [45], it may explainour previous failure to observe intracellular formation ofAGEs, as detected using AGE-antibody [16] in pericytesexposed to high glucose. As reported previously [16, 46,47] intracellular glucose levels were significantly increasedin pericytes exposed to high glucose. During the initialstep of auto-oxidative glycation, ROS fragments glucose togenerate glyoxal with potential to react and form CML onlysine residues of intracellular and extracellular proteins [48].The formation of CML from intermediates of the Mailardreaction, such as Schiff ’s base and the Amadori product thatare raised in pericytes exposed to high glucose [16], requiresROS-mediated oxidative cleavage of the carbon backbone[49]. Although, ROS controls the formation of intracellularCML, its role in glucose-induced apoptosis is unclear, butmay be involved in the observed reduced proliferation andincreased necrosis of pericytes in high glucose. This notion issupported by a recent study showing that CML-modificationof histones is associated with decreased proliferation ofkeratinocytes exposed to glyoxal [50]. Intracellular forma-tion of CML in pericytes could well be important sinceincreased levels of CML are present in the retina of ratswith experimental diabetes [51] and levels in lymphocytesare associated with the pathogenesis of diabetic retinopathy[52].

8 Journal of Ophthalmology

In summary, we have demonstrated for the first timethat in contrast to BAEC, ROS from NADPH oxidase andnot the mitochondria plays a key role in glucose-inducedintracellular formation of CML and apoptosis of retinalcapillary pericytes. The results support the recent reportslinking NADPH oxidase and diabetic retinopathy [39] andthe beneficial effect of antioxidants to prevent oxidativestress and caspase-3-dependent apoptosis of pericytes [19].Specific pharmacological NADPH oxidase inhibitors mightalso be of potential use during the treatment of early diabeticretinopathy.

Nonstandard Abbreviations

ROS: Reactive oxygen speciesBRP: Bovine retinal capillary pericytesBAEC: Bovine aortic endothelial cellsHUVEC: Human umbilical vein endothelial cellsAc-DEVD-pNA: N-acetyl-Asp-Glu-Val p-nitroanilidep-NA: p-nitroanilineFCS: Foetal calf serumPBS: Phosphate buffered salineOH: Hydroxyl radicalH2O2: Hydrogen peroxideO2

−: superoxide radicalBSA: Bovine serum albuminDPI: diphenyleneiodoniumSOD: Superoxide dismutaseDHE: Dihydroethidium hydroethidineETH: EthidiumSDS-PAGE: Sodium dodecyl sulphate

polyacrylamide gel electrophoresisELISA: Enzyme linked immunoabsorbant assayS.E.M: Standard error of meanFAD: Flavin adenine dinucleotideMitoQ: 10-(6′-ubiquinonyl)

decyltriphenylphosphoniumAGE: Advanced glycation product.

Acknowledgments

This paper was supported by the Charitable Foundationand Charities Advisory Trust (London, UK) and Ministryof Science, Technology and Innovation, Malaysia and theForest Research Institute Malaysia (FRIM). Part of this paperwas presented at the EASDEc meeting (abstract published inEur J Ophthalmol 2004 and 2005) and the 8th MichaelsonSymposium on Ocular Circulation and Neovascularization,2005, Ghent, Belgium.

References

[1] D. S. Fong, L. P. Aiello, F. L. Ferris III, and R. Klein, “Diabeticretinopathy,” Diabetes Care, vol. 27, no. 10, pp. 2540–2553,2004.

[2] M. Mizutani, T. S. Kern, and M. Lorenzi, “Accelerated deathof retinal microvascular cells in human and experimentaldiabetic retinopathy,” Journal of Clinical Investigation, vol. 97,no. 12, pp. 2883–2890, 1996.

[3] T. S. Kern, J. Tang, M. Mizutani et al., “Response of capillarycell death to aminoguanidine predicts the development ofretinopathy: comparison of diabetes and galactosemia,” Inves-tigative Ophthalmology and Visual Science, vol. 41, no. 12, pp.3972–3978, 2000.

[4] Diabetes Control and Complications Trial Research Group,“The effect of intensive treatment of diabetes on the develop-ment and progression of long-term complications in insulin-dependent diabetes mellitus,” The New England Journal ofMedicine, vol. 329, no. 14, pp. 977–986, 1993.

[5] UK Prospective Diabetes Study (UKPDS) Group, “Intensiveblood-glucose control with sulphonylureas or insulin com-pared with conventional treatment and risk of complicationsin patients with type 2 diabetes (UKPDS 33),” The Lancet, vol.352, no. 9131, pp. 837–853, 1998.

[6] R. A. Kowluru and P.-S. Chan, “Oxidative stress and diabeticretinopathy,” Experimental Diabetes Research, vol. 2007, pp. 1–12, 2007.

[7] D. M. Niedowicz and D. L. Daleke, “The role of oxidative stressin diabetic complications,” Cell Biochemistry and Biophysics,vol. 43, no. 2, pp. 289–330, 2005.

[8] T. Nishikawa, D. Edelstein, X. L. Du et al., “Normalizingmitochondrial superoxide production blocks three pathwaysof hyperglycaemic damage,” Nature, vol. 404, no. 6779, pp.787–790, 2000.

[9] R. P. Brandes and J. Kreuzer, “Vascular NADPH oxi-dases: molecular mechanisms of activation,” CardiovascularResearch, vol. 65, no. 1, pp. 16–27, 2005.

[10] J. Stolk, T. J. Hiltermann, J. H. Dijkman, and A. J. Verho-even, “Characteristics of the inhibition of NADPH oxidaseactivation in neutrophils by apocynin, a methoxy-substitutedcatechol,” American Journal of Respiratory Cell and MolecularBiology, vol. 11, no. 1, pp. 95–102, 1994.

[11] A. M. James, H. M. Cocheme, and M. P. Murphy, “Mitochon-dria-targeted redox probes as tools in the study of oxidativedamage and ageing,” Mechanisms of Ageing and Development,vol. 126, no. 9, pp. 982–986, 2005.

[12] R. Chibber, P. A. Molinatti, N. Rosatto, B. Lambourne, and E.M. Kohner, “Toxic action of advanced glycation end productson cultured retinal capillary pericytes and endothelial cells:relevance to diabetic retinopathy,” Diabetologia, vol. 40, no. 2,pp. 156–164, 1997.

[13] V. Gurtu, S. R. Kain, and G. Zhang, “Fluorometric andcolorimetric detection of caspase activity associated withapoptosis,” Analytical Biochemistry, vol. 251, no. 1, pp. 98–102,1997.

[14] H. Zhao, J. Joseph, H. M. Fales et al., “Detection andcharacterization of the product of hydroethidine and intra-cellular superoxide by HPLC and limitations of fluorescence,”Proceedings of the National Academy of Sciences of the UnitedStates of America, vol. 102, no. 16, pp. 5727–5732, 2005.

[15] W. O. Carter, P. K. Narayanan, and J. P. Robinson, “Intracel-lular hydrogen peroxide and superoxide anion detection inendothelial cells,” Journal of Leukocyte Biology, vol. 55, no. 2,pp. 253–258, 1994.

[16] R. Chibber, P. A. Molinatti, and E. M. Kohner, “Intracellularprotein glycation in cultured retinal capillary pericytes andendothelial cells exposed to high-glucose concentration,”Cellular and Molecular Biology, vol. 45, no. 1, pp. 47–57, 1999.

[17] E. Beltramo, E. Berrone, S. Buttiglieri, and M. Porta, “Thi-amine and benfotiamine prevent increased apoptosis inendothelial cells and pericytes cultured in high glucose,”Diabetes/Metabolism Research and Reviews, vol. 20, no. 4, pp.330–336, 2004.

Journal of Ophthalmology 9

[18] A. Manea, E. Constantinescu, D. Popov, and M. Raicu, “Chan-ges in oxidative balance in rat pericytes exposed to diabeticconditions,” Journal of Cellular and Molecular Medicine, vol. 8,no. 1, pp. 117–126, 2004.

[19] R. A. Kowluru and P. Koppolu, “Diabetes-induced activationof caspase-3 in retina: effect of antioxidant therapy,” FreeRadical Research, vol. 36, no. 9, pp. 993–999, 2002.

[20] F. Pomero, A. Allione, E. Beltramo et al., “Effects of proteinkinase C inhibition and activation on proliferation andapoptosis of bovine retinal pericytes,” Diabetologia, vol. 46, no.3, pp. 416–419, 2003.

[21] A. M. Abu El-Asrar, L. Dralands, L. Missotten, I. A. Al-Jadaan,and K. Geboes, “Expression of apoptosis markers in the retinasof human subjects with diabetes,” Investigative Ophthalmologyand Visual Science, vol. 45, no. 8, pp. 2760–2766, 2004.

[22] S. Mohr, X. Xi, J. Tang, and T. S. Kern, “Caspase activationin retinas of diabetic and galactosemic mice and diabeticpatients,” Diabetes, vol. 51, no. 4, pp. 1172–1179, 2002.

[23] J. M. Cacicedo, S. Benjachareowong, E. Chou, N. B. Ruder-man, and Y. Ido, “Palmitate-induced apoptosis in cultured bo-vine retinal pericytes: roles of NAD(P)H oxidase, oxidantstress, and ceramide,” Diabetes, vol. 54, no. 6, pp. 1838–1845,2005.

[24] A. Manea, M. Raicu, and M. Simionescu, “Expression of func-tionally phagocyte-type NAD(P)H oxidase in pericytes: effectof angiotensin II and high glucose,” Biology of the Cell, vol. 97,no. 9, pp. 723–734, 2005.

[25] G. Cheng, Z. Cao, X. Xu, E. G. van Meir, and J. D. Lambeth,“Homologs of gp91phox: cloning and tissue expression ofNox3, Nox4, and Nox5,” Gene, vol. 269, no. 1-2, pp. 131–140,2001.

[26] U. Bayraktutan, L. Blayney, and A. M. Shah, “Molecularcharacterization and localization of the NAD(P)H oxidasecomponents gp91-phox and p22-phox in endothelial cells,”Arteriosclerosis, Thrombosis, and Vascular Biology, vol. 20, no.8, pp. 1903–1911, 2000.

[27] J.-M. Li and A. M. Shah, “Intracellular localization andpreassembly of the NADPH oxidase complex in culturedendothelial cells,” Journal of Biological Chemistry, vol. 277, no.22, pp. 19952–19960, 2002.

[28] K. Asaba, A. Tojo, M. L. Onozato et al., “Effects of NADPH oxi-dase inhibitor in diabetic nephropathy,” Kidney International,vol. 67, no. 5, pp. 1890–1898, 2005.

[29] T. Etoh, T. Inoguchi, M. Kakimoto et al., “Increased expressionof NAD(P)H oxidase subunits, NOX4 and p22phox, inthe kidney of streptozotocin-induced diabetic rats and itsreversibity by interventive insulin treatment,” Diabetologia,vol. 46, no. 10, pp. 1428–1437, 2003.

[30] T. J. Guzik, N. E. West, E. Black et al., “Vascular superoxideproduction by NAD(P)H oxidase: association with endothelialdysfunction and clinical risk factors,” Circulation Research, vol.86, no. 9, pp. E85–E90, 2000.

[31] P. Mohanty, W. Hamouda, R. Garg, A. Aljada, H. Ghanim,and P. Dandona, “Glucose challenge stimulates reactive oxygenspecies (ROS) generation by leucocytes,” Journal of ClinicalEndocrinology and Metabolism, vol. 85, no. 8, pp. 2970–2973,2000.

[32] T. Inoguchi, P. Li, F. Umeda et al., “High glucose level andfree fatty acid stimulate reactive oxygen species productionthrough protein kinase C-dependent activation of NAD(P)Hoxidase in cultured vascular cells,” Diabetes, vol. 49, no. 11, pp.1939–1945, 2000.

[33] T. E. DeCoursey and E. Ligeti, “Regulation and termination ofNADPH oxidase activity,” Cellular and Molecular Life Sciences,vol. 62, no. 19-20, pp. 2173–2193, 2005.

[34] A. Fontayne, P. M.-C. Dang, M.-A. Gougerot-Pocidalo, and J.El Benna, “Phosphorylation of p47phox sites by PKC α, βII, δ,and ζ : effect on binding to p22phox and on NADPH oxidaseactivation,” Biochemistry, vol. 41, no. 24, pp. 7743–7750, 2002.

[35] C.-D. Agardh, B. Hultberg, R. C. Nayak, P. Farthing-Nayak,and E. Agardh, “Bovine retinal pericytes are resistant toglucose-induced oxidative stress in vitro,” Antioxidants andRedox Signaling, vol. 7, no. 11-12, pp. 1486–1493, 2005.

[36] N. Ouslimani, J. Peynet, D. Bonnefont-Rousselot, P. Therond,A. Legrand, and J.-L. Beaudeux, “Metformin decreases intra-cellular production of reactive oxygen species in aorticendothelial cells,” Metabolism, vol. 54, no. 6, pp. 829–834,2005.

[37] A. Tawfik, T. Sanders, K. Kahook, S. Akeel, A. Elmarakby,and M. Al-Shabrawey, “Suppression of retinal peroxisomeproliferator-activated receptor gamma in experimental dia-betes and oxygen-induced retinopathy: role of NADPH oxi-dase,” Investigative Ophthalmology & Visual Science, vol. 50,no. 2, pp. 878–884, 2009.

[38] M. Al-Shabrawey, M. Rojas, T. Sanders et al., “Role of NADPHoxidase in retinal vascular inflammation,” Investigative Oph-thalmology and Visual Science, vol. 49, no. 7, pp. 3239–3244,2008.

[39] M. Al-Shabrawey, M. Bartoli, A. B. El-Remessy et al., “Roleof NADPH oxidase and Stat3 in statin-mediated protectionagainst diabetic retinopathy,” Investigative Ophthalmology andVisual Science, vol. 49, no. 7, pp. 3231–3238, 2008.

[40] M. Yano, G. Hasegawa, M. Ish II et al., “Short-term exposureof high glucose concentration induces generation of reactiveoxygen species in endothelial cells: implication for the oxida-tive stress associated with postprandial hyperglycemia,” RedoxReport, vol. 9, no. 2, pp. 111–116, 2004.

[41] A. Tojo, K. Asaba, and M. L. Onozato, “Suppressing renalNADPH oxidase to treat diabetic nephropathy,” Expert Opin-ion on Therapeutic Targets, vol. 11, no. 8, pp. 1011–1018, 2007.

[42] K. Susztak, A. C. Raff, M. Schiffer, and E. P. Bottinger,“Glucose-induced reactive oxygen species cause apoptosis ofpodocytes and podocyte depletion at the onset of diabeticnephropathy,” Diabetes, vol. 55, no. 1, pp. 225–233, 2006.

[43] R. M. Kluck, E. Bossy-Wetzel, D. R. Green, and D. D.Newmeyer, “The release of cytochrome c from mitochondria:a primary site for Bcl- 2 regulation of apoptosis,” Science, vol.275, no. 5303, pp. 1132–1136, 1997.

[44] M. M. Anderson and J. W. Heinecke, “Production ofNε-(carboxymethyl)lysine is impaired in mice deficient inNADPH oxidase: a role for phagocyte-derived oxidants inthe formation of advanced glycation end products duringinflammation,” Diabetes, vol. 52, no. 8, pp. 2137–2143, 2003.

[45] E. D. Schleicher, E. Wagner, and A. G. Nerlich, “Increasedaccumulation of the glycoxidation product N(ε)- (car-boxymethyl)lysine in human tissues in diabetes and aging,”Journal of Clinical Investigation, vol. 99, no. 3, pp. 457–468,1997.

[46] E. Berrone, E. Beltramo, C. Solimine, A. U. Ape, and M. Porta,“Regulation of intracellular glucose and polyol pathway bythiamine and benfotiamine in vascular cells cultured in highglucose,” Journal of Biological Chemistry, vol. 281, no. 14, pp.9307–9313, 2006.

[47] J.-Z. Zhang, L. Gao, M. Widness, X. Xi, and T. S. Kern,“Captopril inhibits glucose accumulation in retinal cells in

10 Journal of Ophthalmology

diabetes,” Investigative Ophthalmology and Visual Science, vol.44, no. 9, pp. 4001–4005, 2003.

[48] R. Nagai, K. Ikeda, T. Higashi et al., “Hydroxyl radical mediatesNε-(carboxymethyl)lysine formation from Amadori product,”Biochemical and Biophysical Research Communications, vol.234, no. 1, pp. 167–172, 1997.

[49] K. J. Wells-Knecht, D. V. Zyzak, J. E. Litchfield, S. R. Thorpe,and J. W. Baynes, “Mechanism of autoxidative glycosylation:identification of glyoxal and arabinose as intermediates in theautoxidative modification of proteins by glucose,” Biochem-istry, vol. 34, no. 11, pp. 3702–3709, 1995.

[50] D. Cervantes-Laurean, M. J. Roberts, E. L. Jacobson, and M. K.Jacobson, “Nuclear proteasome activation and degradation ofcarboxymethylated histones in human keratinocytes followingglyoxal treatment,” Free Radical Biology and Medicine, vol. 38,no. 6, pp. 786–795, 2005.

[51] H.-P. Hammes, A. Alt, T. Niwa et al., “Differential accumu-lation of advanced glycation end products in the course ofdiabetic retinopathy,” Diabetologia, vol. 42, no. 6, pp. 728–736,1999.

[52] H.-P. Hammes, M. Brownlee, J. Lin, E. Schleicher, andR. G. Bretzel, “Diabetic retinopathy risk correlates withintracellular concentrations of the glycoxidation product Nε-(carboxymethyl) lysine independently of glycohaemoglobinconcentrations,” Diabetologia, vol. 42, no. 5, pp. 603–607,1999.